(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "StandardEnvironmentalSystemsInc-OperatingManualRelativeHumidtyTestChamberModelHB-4Serial90079"

STANDARD MICROSYSTEMS 
CORPORATION, 




35 Marcus Blvd Hauppauge NY 11788 
15161273 3100 Fax (5161 231-6004 



PRODUCTS DIVISION 



TECHNICAL NOTE 7-5 




January 1 992 



RS-485 CABLING GUIDELINES FOR THE 
COM20020 UNIVERSAL LOCAL AREA 
NETWORK CONTROLLER (ULANC) 

AND 

EXPERIMENTAL PROCEDURE FOR 
VERIFICATION OF RS-485 
CABLING GUIDELINES 

REVISION D 



By: Shaun Knoll, Systems Engineer and 
Stephanie Colo, Project Engineer 



TABLE OF CONTENTS 



INTRODUCTION TO EIA RS-485 1 

CABLING GUIDELINES FOR RS-485 INTERFACE WITH THE COM20020 1 

TRANSMISSION LINE EFFECTS IN LOCAL AREA NETWORKS 2 

Reflections 2 

Signal Attenuation 4 

D.C. Loading 5 

CALCULATING THE NUMBER OF NODES AND LENGTH OF CABLE .... 8 

BIASING THE NETWORK 9 

CAPACITIVE EFFECTS 11 

CABLING TOPOLOGY AND CONNECTORS 11 

EXPERIMENTAL VERIFICATION OF RS-485 CABLING GUIDELINES 13 

OBJECTIVE 13 

PROCEDURE 13 

SETUP 15 

RESULTS 16 

CONCLUSION 18 

5 Mbs DC COUPLED RS-485 ANALYSIS 19 



INTRODUCTION TO EIA RS-485 



EIA RS-485 is a specification for the support of a multi-drop differential serial digital 
data network. RS-485 came about as microprocessors and the use of distributed 
intelligence became popular in the design of industrial systems. The implementation 
of such concepts created a demand for a standard method of communicating serially 
in such environments. The actual RS-485 specification came about as a result of the 
shortfalls of the RS-422 standard. RS-485 is virtually identical to RS-422 except in 
two respects, increased receiver sensitivity and support of longer line lengths. 

The basic RS-485 specification standardizes the electrical characteristics of each 
transceiver and provides some basic guidelines for establishing a network. By 
definition, a basic transceiver shall have a minimum input resistance of 1 2 Kohms and 
handle a + 1-1 V common mode voltage regardless of whether power is applied or not. 
When the transceiver is powered it must present the minimum input resistance and 
present less than 50pf of capacitance at its input terminals. In addition, each driver 
must be capable of providing a minimum level of 1.5V in the presence of 32 
transceivers and two 1 20 ohm terminating resistors. The 1 20 ohm termination results 
from the use of twisted pair cable, the preferred media in many industrial applications 
because of its wide availability and low cost. Another critical requirement of RS-485 
is that the receiver must be capable of detecting levels down to 200mV which is of 
great advantage when long line lengths are needed. RS-485 does not specify a 
modulation method or a maximum data rate. This gives the system designer great 
flexibility in creating a low-cost high-performance network. 

The combination of long line length, high node count (32 nodes), and support of low- 
cost media has made the RS-485 specification the primary choice as a data 
communication standard in industrial applications. 



CABLING GUIDELINES FOR RS-485 INTERFACE 
WITH THE COM20020 

The following cabling guidelines provide a basis for establishing a low cost Local Area 
Network (LAN) based on the ARCNET protocol for use with the COM20020 Universal 
ARCNET Controller with a differential RS-485 driver. The guidelines presented are for 
unshielded twisted pair cable modulated with the C0M20020's backplane encoding 
scheme. All testing and experiments were performed using a 24AWG copper twisted 
unshielded 2 pair cable with a characteristic impedance (Z ) of 120 Ohms. The 
topology used in all experiments was a daisy-chained configuration with no stubs (i.e. 
no drops). 



1 



TRANSMISSION LINE EFFECTS IN LOCAL AREA NETWORKS 



Transmission line effects often present an obstacle in obtaining high performance in 
data communication networks. Among the problems that plague high data rate LAN's 
are reflections, signal attenuation, and D.C. loading. Taking into account all three 
parameters when designing a network can result in a faster and more reliable network. 

A) REFLECTIONS IN TRANSMISSION LINE 

A reflection in a transmission line is the result of an impedance discontinuity that a 
travelling wave sees as it propagates down the line. To eliminate the presence of 
reflections from the end of the cable you must terminate the line at its characteristic 
impedance by placing a resistor across the line as shown in Figure 1 . 



TERMINATION OF A NETWORK 
Reflections are caused by a discontinuity in the line 

DIRECTION OF PROPAGATION _ 

, , INCIDENT WAVE 




7 = 120 ohm h 



SOURCE (j~L\ REFLECTED WAVE .>R = 50OHM 

Z = 120 OHM 



DIRECTION OF PROPAGATION 



1 Z = 120 ohm I 1 DISCONTINUITY 

PROPERLY TERMINATED NETWORK 

I 1 INCIDENT WAVE 



7 = 120 ohm h 



R = 120 OHM P» ( ri) DRIVER ]>R = 120OHM 

— I 7 = 120 ohm I 

Figure 1 



It is important that the line be terminated at both ends since the direction of 
propagation is bidirectional. In the case of unshielded twisted pair this termination is 
120 ohms. Note that all reflection measurements were made with no stubs on the 
network (i.e. no drops). 

Theoretically, a properly terminated transmission line would produce no reflections at 
all. However in a real network, small reflections are produced since the characteristic 
impedance of the cable cannot be met exactly due to variance in the manufacturing 



2 



process of the cable. Another primary source of reflections is the impedance 
mismatch between a data transceiver and the line, which can cause problems on a 
data network by creating perturbations on the line during an otherwise idle state. 
Reflections affect the network by triggering false transitions (bits) on the line 
receiver's input translating into possible framing errors and CRC errors. 

A measure of the relative strength of the reflection generated by discontinuities along 
the line is called the Reflection Attenuation Factor (RAF). This is a measure of the 
strength of the reflected wave to its incident wave. The RAF can be obtained by 
comparing a reflected wave to its incident wave. The magnitude of the reflected 
wave can be measured by sending a burst of sine waves down a transmission line and 
observing at the sending end the magnitude of the wave after the burst has ended 
(see Figure 2). The reflection measured at the sending end is the reflection generated 
by a discontinuity at the receiving end of the line. The magnitude of the incident 
wave can be measured at the receiving end of the cable, since this is the wave from 
which the reflection is generated. It is important to compensate for line loss when 
measuring the reflected wave because the reflected wave, measured at the sending 
end of the cable, has lost some amplitude due to line loss. 



MEASURING REFLECTIONS IN A NETWORK 

1 Z = 120 Ohm 



Z = 120 Ohm 




2.5Mhz 
SINE WAVE 



10 
NODES 



Z = 120 Ohm" 



R T = 120 



Z = 120 Ohm 



MEASURE REFLECTION HERE 



MEASURE INCIDENT WAVE HERE 



Figure 2 



Measurements were made for twisted pair cable and are summarized in Table 1 . The 
following relationship was used in calculating the Reflection attenuation. 

Reflection attenuation = 20 log (V r8f /V inc ) 



where V ref = reflected voltage (compensated for loss) 

V inc = incident voltage measured at receiving end of line 



3 



Table 1 



Frequency 


312.5 KHz 


625 MHz 


1.25 MHz 


2.5 MHz 


5.0 MHz 


Reflection 
Attenua- 
tion 


-35.12 dB 


-33.19 dB 


-28.89 dB 


-24.52 dB 


-17.83 dB 



These above numbers can be interpreted as follows: 



Assume a +5Vp_ p incident wave at 2.5Mhz, a reflected wave will be generated that 
travels to the incident source at an amplitude of: 



-24.52dB = 0.059 



therefore the reflected wave is 0.059 * 5V = 0.297V. 



In practice, the amplitude of the reflected wave might be smaller because 
discontinuities are generated throughout the line and are not in phase with each other, 
thus providing a canceling effect and decreasing the magnitude of the reflection. 

There are several methods for minimizing the effects of reflections such as squelch 
circuits and D.C. biasing. For the small reflection levels observed during 
experimentation, the recommended choice is D.C. biasing for its simplicity and 
minimum parts count (2 resistors). Biasing the network may cause some duty cycle 
distortion but the ARCNET protocol is insensitive to duty cycle or jitter. The biasing 
network will be discussed later in this guide. 



B) SIGNAL ATTENUATION IN TRANSMISSION LINES 



A second transmission line effect that has a bearing on the performance of LAN's is 
signal attenuation. A transmission line can be modeled as a combination of the 
distributed capacitance of the line, the distributed inductance, and resistance (see 
Figure 3). 



TRANSMISSION LINE MODEL 



R' 

WW 



c 



WW 



R' 



c 



G' 



R' 

WW 



L' 



c 



G" 



WW 

R* 



L' 



FT = UNIT RESISTANCE OF LINE 
L" = UNIT INDUCTANCE 
C = UNIT CAPACITANCE 
G' = UNIT ADMITTANCE 



Figure 3 



4 



The capacitance of the line is formed by the parallel conductor pair. At the distances 
used in LAN's (100's feet), the resistance of the cable is negligible and contributes 
very little to line loss. The majority of line loss comes from the LC combination that 
acts like a low pass filter and tends to attenuate the signal as frequency and distance 
go up. For twisted pair cable, the attenuation rate is given in Table 2. These are 
measured values. 



Table 2 - Signal Attenuation 



Frequency 


312.5 KHz 


625 KHz 


1.25 MHz 


2.5 MHz 


5.0 MHz 


Attenuation 
per 100ft. 


-0.4 dB 


-0.6 dB 


-1.0 dB 


-1.3 dB 


-2.0 dB 



C) D.C. LOADING IN RS-485 NETWORKS 



The third parameter that affects network performance is D.C. loading. The D.C. load 
presented to a line driver is a combination of three parameters: 

1) loading effect of termination resistors 

2) loading effect of biasing resistors 

3) loading effect of RS-485 transceivers 

EIA RS-485 specifies that a line driver must be capable of presenting a 1.5V signal 
differentially at its outputs under the loading of 32 receivers and two 120 ohm 
termination resistors. Each receiver or passive transceiver is to provide a minimum 
input impedance of 12K ohms. The total parallel combination load impedance is 51 
ohms, which includes the receiver load, termination resistors, and biasing resistors. 
Since the worst-case specification is restrictive, it may not be practical to design to 
this worst-case because, quite often the typical RS-485 transceiver can drive 
substantially more than the worst-case load. Typically, many RS-485 drivers can 
drive quite a bit more than the 51 ohms, sometimes as low as 20 ohms. If typical 
characteristics are used, a network can be composed of many more nodes than the 
32 specified by EIA RS-485. From laboratory experience, SMC recommends using 
typical transceiver characteristics when determining the D.C. load. In very high 
reliability applications (i.e. medical electronics, avionics), SMC recommends the use 
of the worst case RS-485 parameters. 

The D.C. biasing of the network is essential to providing reliable operation at high 
data rates. The D.C. bias offsets the NULL voltage of the network when all RS-485 
transceivers are in their TRI-STATE mode. A small offset is needed because small 
reflections will occur that will cause spurious transitions on the RS-485 receivers 
input. This would cause the C0M20020 to receive an undesired extra bit, thus 
causing framing or CRC errors and corrupting the data. It is only the negative half of 
the reflected wave that is of concern, so the effective reflection is .1 5V. The largest 
reflections seen on the network are approximately 0.3Vp_ p differentially. This is more 
than enough voltage to trigger a transition on the receiver's input since the receiver 



5 



has a hysteresis of 50mV and is thus quite sensitive (200mV differential sensitivity). 
By installing the biasing network of Figure 4 at each node, the network will be offset 
enough to keep the network in an inactive state. The + 0.35V differential offset 
provided, will keep the line in an inactive state even in the presence of a 0.3Vp_ p 
reflection (see Figure 5) which could cause a false bit to be recognized by the 20020 
and cause data rejection by the receiving node. SMC recommends a pull-up and pull- 
down resistor (see Table 3 for correct value) in the circuit diagram shown in Figure 4. 



RS-485 
TRANSCEIVER 



BIASING NETWORK 
+5V 



R1 




-a- 



Hl 



-a 



LO 



TO NETWORK 



R2 



Figure 4 



If a larger offset is desired, the offset voltage can be calculated as follows: 

To calculate the correct biasing voltage, the maximum reflection should be 
known: 

From laboratory measurements maximum V ref = 0.3Vp_ p . We are only 
concerned with the negative portion of V rof = 0.15V 



so 



l ro< <= 0.15V/(R t1 |j R t2 ) 



l ref < = 2.5ma 



where R,, = R t2 = 1 20 ohms = termination resistance 
and R t = R t1 | j R t2 



We want (1^,, - IfJR, > = 50mV minimum 



6 



The 50mV level is the hysteresis value for many RS-485 transceivers 
therefore 1^., > = 3.33ma 

To get the proper bias resistor values the following relationship can be used: 

+ 5V = lbi«.(Rpull- up + ^pulldown + (Rtl II ^t2^ 

This yields a bias resistor value of 720 ohms for the entire network, 
which gives a bias voltage of 0.35V. 



EFFECTS OF BIAS ON REFLECTIONS 



+V 
ov 

-V 



+v 



DIFFERENTIAL SIGNAL 

W/ NO BIAS 

DRIVER TRI-STATES 



REFLECTION OCCURS (- 3V ) 
CAUSING FALSE TRANSITION 

DIFFERENTIAL SIGNAL 

W/ 720 OHM BIAS RESISTORS 
DRIVER TRI-STATES 

.35V OFFSET 



RAISES REFLECTION TO .1V 



REFLECTION OCCURS 
Figure 5 



(•3V) 



In practice, it is better to provide a larger bias resistor at each node so that the parallel 
combination of all nodes can provide a proper offset voltage. A lower resistor value 



7 



than the suggested may be used if a large amount of noise is present in the system 
or if larger reflections are encountered. It should be noted, that the biasing 
arrangement adds to the total D.C. loading of the network, and yields a total load of 

Rtoul = Rt1 ! i Rt2 I ! (Rrcvr/N) 1 1 (R bia8 /N) 

where R t1 = R t2 = termination resistance 

R rcvr = receiver impedance 
Rbi«i = bias resistance 
N = Number of Nodes 



CALCULATING THE NUMBER OF NODES AND LENGTH OF CABLE 

Three parameters must be taken into account when deciding upon the configuration 
of the network (i.e. line length and number of nodes). These parameters are: 

1) D.C. load 

2) Cable attenuation 

3) Noise margin 

The first two parameters have been previously discussed and the third parameter, 
noise margin, will be addressed now. The noise margin is the minimum voltage above 
the 200mV receiver sensitivity limit prescribed by EIA RS-485. All further calculations 
will assume a OV noise margin. Your individual application may require a greater noise 
margin then that prescribed by the EIA RS-485 specification. The following 
relationship may provide 

= .8(V driver - V ta - V nob . - VaJ 

where V.^ = voltage at end of line 

v driver = driver output voltage 

Vict = voltage loss due to cable 

V no«. - noise margin 

V bi« = D.C. bias voltage applied to network at each node 
(typically .4V) 

.8 = Derating for cable tolerances (± 20%) 



8 



The driver output voltage is a function of the D.C. load presented to the driver by the 
network and can be calculated as described above. For convenience, Table 4 contains 
the number of nodes vs. D.C. load using the typical transceiver characteristics for the 
751 76B RS-485 transceiver. 

V lo „ can be found by determining the driver output voltage for a given load and using 
Table 2 to calculate the line loss for a given distance and frequency. 

Usually, either the number of nodes required or the maximum cable distance are 
known prior to designing a network. By using the above relationship all the remaining 
variables (i.e. number of nodes or maximum cable length) can be determined and 
combined to implement a working network. 



BIASING THE NETWORK 

From experimental results, a good value for biasing resistors is 810 ohms. This value 
can be implemented in two ways. One way is to provide one set of resistors for the 
entire network with a separate power supply for the bias. The second option is to 
provide separate bias resistors at each node, thus providing greater flexibility in adding 
nodes and increased reliability. This has the effect of increasing bias as nodes are 
added, automatically compensating for the additional reflections generated, but 
decreasing the dynamic range due to loading. SMC recommends the use of this 
second option. Table 3 shows the recommended values for different numbers of 
nodes. Note that no one value is optimal because as the node numbers change the 
loading of the network will increase due to biasing. 

Table 3 - Bias Resistance vs. Number of Nodes for 2.5 Mbs 



Number of Nodes 


Bias Resistance 


1 - 10 


2.7K 


11-20 


12.0K 


21 - 30 


18.0K 


31 - 40 


27.0K 



Table 4 contains the estimated load and corresponding driver output voltage based on 
the typical characteristics of the 75176B transceiver and the bias resistance from 
Table 5. Note that these are typical characteristics and your driver may vary 
significantly from these numbers. 



9 



Table 4 - D.C. Load Table 



Number of 
Nodes 


Total Load 
(Ohms) 


Estimated V OUT 
(Volts) 


1 


57.41 


2.5 


5 


48.97 


2.4 


10 


48.70 


2.4 


15 


52.17 


2.4 


20 


50.00 


2.4 


25 


49 65 


2.4 


30 


48.00 


2.4 


35 


47.89 


2.4 


40 


46.55 


2.3 



Table 5 takes into account line loss and shows Table 4 with the maximum distance 
that the network can be driven using the various data rates supported by the 
COM20020 (i.e. 312.5Kbs, 625Kbs, 1.25Mbs, 2.5Mbs). A 0.1 volt noise margin 
was used and a 0.200V receiving end voltage were used in establishing the criteria 
for cable length. The maximum distance driven is relatively independent of load for 
the node numbers of interest (<100). This is evidenced by examining Table 4 and 
noticing that the output voltage does not vary significantly as the number of nodes 
increases. Therefore, Table 5 shows the maximum distance using a driver output of 
2.4V. 



Table 5 - Line Len 


gth vs. Maximum Number of Nodes 


Maximum 
Number of 
Nodes 


Maximum 
Distance 
@312.5 KHz 


Maximum 
Distance 
@625 KHz 


Maximum 
Distance 
@1.25 MHz 


Maximum 
Distance 
@2.5 MBs 


21 


2900 ft. 
(880 m) 


1940 ft. 
(590 m) 


1160 ft. 
(350 m) 


900 ft. 
(270 m) 



It should be noted that these figures are for the typical characteristics and represents 
a nominal case. The worst case output voltage for a RS-485 transceiver is 1.5V 
minimum. Typically, this worst case driver output is only encountered under heavy 
load conditions. If your system exhibits this parameter, measurements should be 
taken to correct for the lower driver output. A lower V driver will result in shorter line 
lengths, however, if longer line length is needed less nodes can be accommodated. 



10 



CAPACITIVE EFFECTS 



Line capacitance can have a significant effect on the performance of the network. 
The loading effect of the capacitance will affect both line length and the number of 
nodes. 

Line capacitance is a combination of two sources. The first source is the effective 
capacitance of the two twisted wires. This is the dominant source and is directly 
proportional to line length (i.e as length increases so does C, ina ). The second source 
is attributed to the capacitor formed from one wire to the insulation. This secondary 
capacitance is often substantially smaller but cannot be ignored. In twisted shielded 
cables, there is additional capacitance from conductor to the shield which can be quite 
appreciable in some cable types. 

Network performance is often degraded by line capacitance and depends on the 
numbers of logical one's and zero's in a particular byte. Additionally, significant duty 
cycle distortion can occur due to how much the line capacitance can charge before 
it is discharged. For example, if a data byte of value 01 h is transmitted across a cable 
in an RS-485 network, a significant amount of time is spent with the line voltage at 
+ 5V differentially. This amount of time is enough for the line to fully charge. When 
the logical one is driven (OV), the line must be completely discharged in order for the 
voltage to fully reach is lowest level. Quite often in high data rate networks, the bit 
time is much less than the time constant of the line. This will cause the duty cycle 
of the bits to vary depending on the bit rate, line length, and data pattern. The more 
logical one's in a given byte the better the duty cycle, because the line never gets to 
fully charge due to the many one's that are being transmitted. A secondary effect of 
line capacitance is bit jitter and is related to duty cycle distortion. Bit jitter results in 
varying pulse widths being received at the RX input of the COM20020 device. If the 
pulse width falls below 10ns then the bit is not detected by the COM20020 resulting 
in framing and CRC errors. 

If capacitive effects are significantly degrading a network two approaches can be 
taken to resolve the problem. Solution one is to decrease the data rate of the network 
so as to reduce the duty cycle distortion and bit jitter, and the second is to use a 
special low capacitance cable specially designed for data communications. This type 
of cable often has increased costs as compared to telephone grade cable and may not 
be suitable for your every application. 



CABLING TOPOLOGY AND CONNECTORS 

SMC recommends a daisy-chained wiring scheme to reduce the reflections caused by 
long stubs used in a trunk-drop topology. RJ-1 1 connectors should be used through 
the network with a 1:1 wiring scheme (no reverse wiring). TEE connectors such as 
MOD-TAP p/n S09-300-444 can be used for wiring convenience but the stub should 
be kept as short as possible. If connectors other than RJ-1 1 are used measurements 



11 



should be taken to insure that there is no insertion loss due to the connector. If 
insertion loss exists, the following equation can be used to calculate line length: 

V = N/V - V, - V - V • - Vu: - (V- * N) 

rcvr driver " I cms noise * margin * bias ' insertion 1 ' ' 

where V rcvr = voltage received at the end of the cable 

Vic. = voltage loss from cable attenuation 

Vow = driver output under D.C. load of network 

V noi.e = noise margin 

Vmrom = tolerance of cable ( + /- 20%) 

Vbia. = Bias voltage applied to the network (typically .4v) 

Vinsertion = insertion loss 

N = number of nodes 

There are several manufacturers of unshielded twisted pair cable. Here are a few 
vendors of this type of cable: 



Vendor 


Cable Number 


BELDEN 


9562 


ALPHA 


55262 


MOD-TAP 


S39-102 



For lengths greater than those listed in the above tables, SMC recommends inserting 
RS-485 repeaters in-line or using a hub in a Star topology. Note that, when using a 
hub or repeater, the device will act as one node. Be sure to take this into account 
when designing the network. 



DAISY-CHAINED TOPOLOGY 



NODEN 



NODE N + 1 



Figure 6 



NODE N+2 



12 



ADDENDUM I 



EXPERIMENTAL VERIFICATION OF RS-485 CABLING GUIDELINES FOR 

THE COM20020 



OBJECTIVE: 

To establish a correlation between Distance and Number of Nodes (based on the 
theory presented in the Cabling Guidelines) by observing network operation when 
distance is varied and when the number of nodes is varied. The data rate is held 
constant at 2.5 Mbps. Robust network operation is based on the following criteria: 

a. Reflections should never cause the idle level to drop below -0.050 V, 
measured differentially. 

b. All negative active levels should be more negative than -0.200 V, 
measured differentially. 

c. Pulse widths of RXIN signals on various nodes should be greater than 
100 nsec. 

d. A thorough transmit/receive test (denoted here as Ping Pong) should 
operate successfully between the farthest nodes on the network and all 
nodes in between. 

The media used is 24 AWG solid copper unshielded 2-pair twisted pair with 
characteristic impedance of 120 Q. The network is configured in a stand-alone bus 
topology containing no repeaters. The cable is terminated using 120 Q resistors on 
either end of the bus. The drops used to attach the nodes to the bus are MOD-TAP 
part# S09-300-444, which have a drop length of 3 in. and negligible insertion loss. 
The RS-485 transceivers used are National's 751 76B. If a different physical media 
is to be evaluated, the criteria and methodology of this document may be followed. 
If the usage of repeaters are to be evaluated, bit jitter could be aggravated, and it is 
recommended to tighten the requirements on RXIN pulse width to approximately 1 80 
nsec in order to protect against bit jitter error. 



PROCEDURE: 

1. Configure the network to contain the maximum distance of cable specified in 
the Cabling Guidelines (900 feet). Connect the maximum number of nodes to 
the network which allows the criteria described above to be met. Attaching the 
appropriate bias resistors to the nodes fights reflections but hinders the 
minimum negative active level. Begin with 2.7 KQ bias resistors on each node. 



13 



To determine whether the measurements fall within the criteria, take 
measurements as follows: 

a. To differentially measure reflections, connect Channels 1 and 2 of the 
oscilloscope to the cable interface pins of the RS485 transceiver (pins 
6 and 7 on the 75LS176). Observe the signals differentially by setting 
the oscilloscope to ADD and INVERT Channel 2. Connect the ground 
lines of the scope probes together; do not connect to system ground. 
Observe the tokens sent by each node to ensure that the idle level never 
drops below -0.050V, measured differentially. Pay special attention to 
the area just following the second DID of the token. 

b. To differentially measure the minimum negative active levels, maintain 
the same setup as in (a). Observe the tokens sent by each node to 
ensure that negative active levels are always more negative than - 
0.200V, measured differentially. Pay special attention to the token sent 
by the furthest physical node from the one being observed. This token 
should appear to have the most attenuation. 

c. To measure the RXIN pulse width, place Channel 1 on the RXIN signal 
of the COM20020 (pin 17 of the DIP package, pin 20 of the PLCC). 
View Channel 1; this is not a differential measurement. It may help to 
see each desired transmitted token by triggering externally off of the 
TXEN of the transmitting node. Ensure that all RXIN pulses are present 
and that they are all greater than 100 nsec wide. Due to the biasing of 
the network, the capacitance of the cable, and the RS-485 transceivers, 
the pulses most likely to disappear are the negative transitions (logic 
"1 "s to the COM20020) immediately following a long duration of a high 
level (many logic "0 n s to the COM20020). Pay special attention, for 
instance, to the negative transition in the 04H pattern (End Of 
Transmission pattern) and the negative transition immediately following 
this pattern. Again, the token sent by the furthest node is most likely to 
have problems as viewed by the observed node. Although the 
COM20020 data sheet specifies a minimum RXIN width of 10 nsec, it 
is important that the RXIN signals be greater than 100 nsec. This is 
because bit jitter, which is affected by duty cycle distortion due to the 
capacitance of the network, must be taken into account. As long as the 
RXIN width is greater than 100 nsec, it is guaranteed that bit jitter will 
not cause the receive circuitry of the COM20020 to lose 
synchronization. 

2. Record the results obtained in Step 1, including the worst reflections on the 
network, minimum negative active level of the worst transition from the 
furthest node, RXIN width of the worst transition from the furthest node, and 
whether the Ping Pong test works between all nodes on network. 



14 



3. Detach 100 feet of cable, keep two nodes attached to network, and make the 
same measurements as in Step 1 and record. 

4. Add nodes to the network, remeasuring the parameters, until the 
measurements are similar to those obtained previously (worst reflection 
amplitude, minimum negative active level, RXIN width, and behavior of Ping 
Pong). 

5. Repeat steps (3) and (4) until Distance vs. Number of Nodes curve is complete. 



SETUP: 



See Figure 1 for the setup of the experiment. The media used is 24 AWG solid 
copper unshielded 2-pair twisted pair with characteristic impedance of 120 CI. The 
network is configured in a stand-alone bus topology. The cable is terminated using 
120 Q resistors on either end of the bus. The drops used to attach the nodes to the 
bus are MOD-TAP part# S09-300-444, which have a drop length of 3 in. and 
negligible insertion loss. The RS-485 transceivers used are National's 751 76B. 



P.C. Node 




20020 




20020 




20020 




20020 




20020 




20020 


NID-FD 




NODE 




NODE 




NODE 




NODE 




NODE 




NODE 




20020 
MO-FA 



20020 
NODE 



20020 
NODE 



20020 
NODE 



20020 
NODE 



20020 
NODE 



20020 
NOOE 



CH 1 
+ CH2 
CH 3 




Scope views Minimum Negative Active Level of 
Token passed by NIO FD by adding and Inverting 
Channel 2. FDQN width Is viewed on Channel 3. 



Figure 1 



15 



RESULTS: 



Table 1 - Results 
(Each node containing 2.7 KC1 bias resistors) 



Configuration 
(#Nodes:Feet) 


Minimum 
Idle Level 
(including 
reflections) 
(rolls) 


Minimum 
Negative 
Active 
Level 
(volts) 


RXIN 
Width 

(nsec) 


Ping 
Pons 


No. of Additional Nodes to 
Bring to Minimum Negative 
Active Levels Previously 
Obtained at: 
olMI/7UU/oUO/5HMl r eet 


2:900 


+0.050 


-0.300 


100 


Yes 

1 CO 


/ / / 


2:800 


+0.050 


-0.500 


120 


Yes 


-/-/-/4 


2:700 


+0.100 


-0.600 


140 


Yes 


-/-/4/10 


2:600 


+0.100 


-0.700 


160 


Yes 


-/5/10/14 


2:500 


+0.100 


-0.800 


180 


Yes 


5/10/14/* 



1 . The Cabling Guidelines specify that 900 feet is the maximum distance for a 
network. Experimentally, it was found that only 2 nodes may exist at 900 feet 
to safely meet the criteria listed above. 

2. 100 feet of cable was detached. At 800 feet, the addition of 4 more nodes 
was required to bring the Minimum Negative Active Level to those obtained for 
2 Nodes at 900 feet. 

3. Another 100 feet of cable was detached. At 700 feet, the addition of 4 more 
nodes was required to bring the Minimum Negative Active Level to those 
obtained for 2 Nodes at 800 feet, and the addition of 10 for those obtained at 
900 feet. 

4. Another 100 feet of cable was detached. At 600 feet, the addition of 5 more 
nodes was required to bring the Minimum Negative Active Level to those 
obtained for 2 Nodes at 700 feet, the addition of 10 for those obtained at 800 
feet, and the addition of 14 for those obtained at 900 feet. 

5. A final 100 feet of cable was detached. At 500 feet, the addition of 5 more 
nodes was required to bring the Minimum Negative Active Level to those 
obtained for 2 Nodes at 600 feet, the addition of 10 for those obtained at 700 
feet, and the addition of 14 for those obtained at 800. * When 18 additional 
nodes were added, the minimum negative active level was similar to that 
obtained for 800 feet, which means that as cable distance decreases, the 
number of nodes begins to increase more rapidly. 



16 



20 



18 



16 



14 



Total 12 
No. 
Nodes 10 



;Fach"npcTe 
reontainiing- 
_;blasJ3'$L 



Each node 
^containing 



bias R's 



of 12KOhm 



;of 2.7 kOhm 



100 200 300 400 500 600 700 800 900 

Distance 
(feet) 



Figure 2 



17 



Note that the previous values were all obtained using bias resistors of 2.7 KQ. Once 
the node count exceeds 10, it is suggested to place higher value bias resistors on 
each node. This brings the bias level down to a reasonable level, allowing the 
minimum negative active level to become higher, but keeping the level of reflections 
still within specification. 

The Distance vs. Number of Nodes curve, obtained from the previous results (using 
2.7 KQ bias resistors), is presented in Figure 2. Also in Figure 2 is a portion of the 
curve which results in placing 12 KQ bias resistors on each node. 

The following are additional results obtained, which may be helpful to the user, 
including the Edge Rate (around OV differential) of the worst pulse seen from 20 
unique tokens and the rise and fall time of a typical pulse. They are presented here: 

Rising Transition Edge Rate = 22.2 Volts/^sec worst case. 
Falling Transition Edge Rate = 7.27 Volts/^sec worst case. 

Rise Time = 280 nsec typical. 
Fall Time = 130 nsec typical. 



CONCLUSION: 

It was seen that the curve was relatively linear when one bias resistor value is placed 
on each node. Since the slope of the curve is approximately 4 nodes for every 100 
feet (in terms of Minimum Negative Active Level), we can see that the addition of a 
1 00 foot segment of cable has four times the impact versus the addition of one node. 
Therefore, the addition of two more nodes at 600 feet does not significantly hinder 
the network performance. In fact, connecting a total of 18 nodes at 600 feet with 
2.7 KO bias resistors still showed perfect Ping Pong operation between all nodes. In 
addition, 10 Nodes at 1000 feet was observed, with the following results: 
Reflections of -0.100 V, a minimum negative active level of -0.200 V, and an RXIN 
width of 40 nsec. Ping Pong operation was adequate, but measurements such as 
these should be considered marginal because pushing the limits of the curve can 
represent marginal behavior. At any time, a tradeoff exists between reflections and 
the minimum negative active level. Adding many nodes to the network increases the 
probability of significant reflections. To overcome this, one might be inclined to raise 
the bias of the network. At some point, however, the minimum negative active level 
will no longer fall within the specified limits of the RS-485 Receiver. 

Alternatively, increasing the number of nodes increases the total bias of the network. 
Therefore, to keep the minimum negative active level within specification, the bias 
resistor of each node may be increased thus decreasing the overall bias of the 
network. Increasing the bias resistors on each node to 1 2 KQ allowed a large number 
of nodes to operate over a longer distance. However, please note that reflections are 
more likely to cross the specified threshold. The results of this experiment indicate 
that 14 nodes at 900 feet, or 19 nodes at 800 feet exhibited successful results. No 
additional testing was performed above 19 nodes. 



18 



ADDENDUM II 



5 Mbs DC COUPLED RS-485 ANALYSIS 

As a result of future customer requirements, an analytical and experimental analysis 
was performed with ARCNET running at a 5Mbs data rate using an RS-485 
compatible physical layer. This analysis will present line loss, reflection values, 
biasing concerns, and a distance vs. load chart. 

LINE LOSS 

Using a symmetrical 5Mhz sine wave burst, the average line loss for UTP with 120 
ohm terminations was found to be -2.0dB per 100 feet. 

REFLECTIONS AND RAF 

From experimental results, the maximum reflection measured was.50Vp. p . The 
average RAF was approximately -15.0dB of the incident wave. 

DC BIASING 

As discussed in TN 7-5, DC biasing of the network is necessary to avoid unwanted 
transitions as a result of reflections generated along the line. Although the maximum 
reflection is .5 volts, we are only concerned with the negative portion of the wave, 
so for calculation of the bias resistors a value of .25V will be used. Due to the larger 
reflections generated at 5Mbs, the following chart should be used: 



NUMBER OF NODES 


RESISTOR VALUE 


2 - 5 


1.8K 


6- 10 


2.7K 


11-20 


6.8K 


21 - 32 


12.0K 



Note: One set of bias resistors can be used for the entire network. In this case, a 
value of 470 ohms should be used. 

ANALYTICAL PERFORMANCE 

Analytical performance is based on several measured parameters including line loss 
and maximum reflections given a 5Vp. p input. The bias resistors calculated above 
were determined based on the maximum reflection. If a smaller incident is to be used, 
the maximum reflection can be calculated using the RAF. Three cases are considered 
in the analysis: a nominal case in which the 485 transceiver outputs a + 5V signal; 



19 



a typical case where the driver outputs 2.4V; and a worst case where the driver 
outputs 1.5V. 



NUMBER OF 

NODbb 


NOMINAL 
(5 V) 


TYPICAL 
(2.4V) 


WORST CASE 
(15V) 


1 - 5 


1 IKft 


aooft 




6 - 10 


1 .IKft 


800ft 


550ft 


11-20 


1 .1 Kft 


800ft 


550ft 


21 - 32 


I.IKft 


775ft 


525ft 



Note: The above numbers are calculated and do not take into account bit jitter or 
duty cycle distortion effects. 



MEASURED PERFORMANCE 

The procedure and criteria used for determining acceptable performance are identical 
to those outlined in Addendum I (Experimental Verification of RS-485 Cabling 
Guidelines for the COM20020). The only parameter that was changed was the bit 
jitter budget, which was decreased from 100ns to 50ns due to the change in clock 
speed. The following results were obtained: 

Maximum distance - 600ft Maximum number of nodes - 17 

The maximum distance achieved was found with a +5v driver output. The primary 
reason for the decreased performance is attributed to duty cycle distortion as a result 
of the biasing network. 



20